The Room Between Sessions
The second baseline measurement took forty minutes.
Sonmat-4471 stood in each corner of room 14 in sequence, held the depth-gradient instrument at waist height, and waited for the reading to stabilize. Northeast: 0.44. Northwest: 0.39. Southeast: 0.52. Southwest: 0.48. She took each reading twice, confirmed the second matched the first within tolerance, and recorded them in the measurement log.
Then she sat down on the floor of the room with the log open in her lap and looked at the numbers.
The first baseline had been 0.42, 0.38, 0.51, 0.47. The second baseline was 0.44, 0.39, 0.52, 0.48. Every corner had increased by 0.01 or 0.02 over forty-eight hours. The spatial distribution was unchanged: the southeast corner was still the highest, the northwest still the lowest, by the same margin as before. Whatever was generating the southeast-high pattern was stable. The overall level of the gradient was rising.
In the forty-eight hours between the first and second measurements, no sessions had taken place in room 14. The building coordinator had confirmed this. The room had been locked; she had the only key card for the week. Nothing had happened in room 14 since she had left it after the first baseline measurement.
She had designed the protocol assuming the gradient would be stable.
The protocol she had designed assumed the baseline would be stable — that the room's accumulated depth would represent a fixed quantity, a static reservoir she could measure and then use as a control in the source-session experiment. She had designed the three-session baseline sequence precisely to confirm stability: if the baseline was the same across three measurements, she could treat it as a known quantity. If it varied, she had to account for the variation before she could interpret any source-session effects.
She had expected it to be stable. She was sitting on the floor of room 14 because it was not.
There were three explanations.
The first was measurement error. The instrument had a stated tolerance of plus or minus 0.02 units. Both corners that had increased by 0.01 were within tolerance. The two corners that had increased by 0.02 were at the edge of tolerance. If the instrument's uncertainty was distributed toward the high end of its tolerance range, both measurements could be consistent with the same underlying reality.
She did not think this was the explanation. She had confirmed each reading twice, and the two readings had agreed. The instrument was consistent with itself, which was not a guarantee of accuracy but was evidence against random error. The pattern — every corner rising, by small and consistent amounts, with the spatial distribution preserved — looked more like a real change than like a uniform shift in measurement conditions.
The second explanation was environmental. The room's temperature or humidity or electromagnetic environment had changed between measurements in a way that affected the gradient instrument's readings without reflecting any change in the actual depth gradient. She would need to control for environmental variables in the third measurement, which she had not done in the first two.
She added this to the protocol note for the third baseline: measure temperature, humidity, and ambient electromagnetic field in each corner alongside the gradient reading.
The third explanation was that the room's depth gradient was actually increasing.
She sat with this for a while.
If the gradient was increasing without active sessions — if the room, unoccupied and locked, was accumulating depth from some source she had not identified — then several things she had assumed about how depth worked were wrong. She had assumed that depth accumulated through sessions: through practitioners and patients doing the work of felt-capture in a space over time. She had assumed that a room without sessions was not accumulating anything. The protocol she had designed treated the locked, unused room as inert.
The second baseline suggested it was not inert.
She thought about what sources of depth a locked room might have that she had not accounted for. The building itself was running its ordinary infrastructure: the relay junction on the floor below, the ventilation system cycling through the afternoon, the building's structural systems doing whatever they did continuously. Practitioners walked past the door. Sessions were running in the rooms on either side.
She stopped.
She had not thought about the adjacent rooms.
Room 14 was between rooms 13 and 15, both of which were active session rooms currently in use. She had checked the building schedule when she booked room 14. Rooms 13 and 15 had sessions today. They had sessions two days ago when she had taken the first baseline. They had sessions most days of the week. She had chosen room 14 as a storage-only space without ongoing session activity, and she had been correct about that. She had not considered that the adjacent spaces were running sessions continuously.
If depth bled through walls — which was precisely what her experiment was designed to test — then room 14 was receiving depth from rooms 13 and 15 every time either of those rooms held a session. The locked room was not isolated. It was adjacent.
She was not sure whether to be frustrated or interested.
Both, she decided. The protocol had a design flaw she had not seen when she designed it. The experiment's premise — that room 14 could serve as a controlled baseline space — was wrong if the adjacent-room effect existed, because the adjacent-room effect was exactly what she was trying to measure, and it would be operating on her baseline during the measurement period.
There was a way to read this as a validation of the hypothesis. If the adjacent rooms were causing room 14's gradient to increase, then the adjacent-room effect was real and she was detecting it. The experiment was working. The baseline was detecting precisely the effect she had designed the source session to produce.
There was also a way to read this as a confound she had designed into the experiment and would need to control for before any of her results were interpretable.
She needed the third baseline and the environmental controls and a floor plan showing which rooms had sessions and when before she could tell which reading was correct.
She stood up from the floor of room 14, replaced the key card in her pocket, and wrote three items in the measurement log: get floor plan, check schedule for rooms 13 and 15, add environmental controls to third baseline. Then she unlocked the door and went to find the building coordinator.
The corridor outside was quiet. One floor below, the relay junction was doing its midafternoon work. The building was occupied and running, as it had been every day for forty years. Room 14 was sitting in the middle of it, locked, accumulating.
The building coordinator was at her desk at the end of the ground-floor corridor, across from the building's main information board, which had the week's session schedule posted in the usual format: room numbers, times, practitioners, session types. Sonmat stood in front of it and checked rooms 13 and 15.
Room 13: sessions Tuesday and Thursday 10 AM–1 PM, practitioner Hye-bin Cho. Today was Thursday. Room 13 had been running a session for the last two and a half hours.
Room 15: sessions Monday, Wednesday, Friday 9 AM–12 PM and 2 PM–5 PM. Today was Thursday. Room 15 was quiet today.
She ran the numbers.
First baseline, 48 hours ago: room 13 — Tuesday session complete (had ended before she took the measurement); room 15 — Wednesday sessions complete (both had ended before she took the measurement). Room 14 after both adjacent rooms' last sessions.
Second baseline, now: room 13 — Thursday session in progress (had been running for two and a half hours when she took the measurement). Room 15 — quiet.
The second baseline had been taken while room 13 was actively running a session. If the adjacent-room effect was real, room 14's gradient at the time of the second baseline would reflect the influence of an ongoing session in room 13, not just the residual from completed sessions as in the first baseline.
This was not a flaw in the protocol. This was information.
If the source of the gradient increase was room 13's ongoing session, the increase would be concentrated in the northeast and northwest corners of room 14 — the corners closest to the shared wall with room 13. She checked the first and second baseline numbers again.
NE: 0.42 → 0.44. Increase of 0.02. NW: 0.38 → 0.39. Increase of 0.01. SE: 0.51 → 0.52. Increase of 0.01. SW: 0.47 → 0.48. Increase of 0.01.
The largest increase was in the NE corner, adjacent to the room 13 wall. The second-largest was in the SE corner, which was farther from room 13 but adjacent to the southeast-wall baseline elevation she had noted from the first measurement — that corner was already the highest in the room and had increased by the same amount as the far corners.
The pattern was not definitive. But the northeast corner's larger increase, in the corner closest to room 13, was consistent with the adjacent-room hypothesis.
She sat down in one of the building coordinator's waiting chairs and wrote in the measurement log for another ten minutes.
Then she asked the coordinator: when does room 13 typically have its first session of the week?
Tuesday, the coordinator said. Ten AM.
She asked: is there any period when both rooms 13 and 15 are consistently quiet for several days?
Not typically, the coordinator said. There might be a weekend break if the practitioners are away, but the building stays active through most weekends.
Sonmat thanked her and left.
The experiment had been designed to control for the adjacent-room effect by choosing a storage room. Instead, she had found the adjacent-room effect operating during her baseline period, which was either a catastrophic design flaw or the best possible thing that could have happened to the experiment.
She walked back to her work area and began revising the protocol.
The third baseline, originally scheduled for 96 hours after the second, would be taken when room 13 was quiet — Monday morning, before the Tuesday sessions began. She would also take a fourth baseline 48 hours after that, when room 13 had run two sessions since the third baseline. The comparison would give her the before-and-after picture of the adjacent room's direct effect on room 14's gradient.
This was a better experiment than the one she had designed.
She was not pleased that it was better because the original design was flawed. She was pleased that she had found out early enough to fix it.
She looked at the revised protocol in the notebook for a long time. The original experiment had been: source session in room 14, measurement in room 14 during and after, compare to baseline. The revised experiment was the same source session plus a controlled comparison of room 14's gradient before and after two adjacent sessions in room 13 — with the source session run in room 14 after that comparison was established.
The study had grown. It had grown in the right direction, because the adjacent-room comparison was directly relevant to the hypothesis. But it would take longer, and the timing constraints were tighter: she needed the Monday baseline, then the post-Tuesday-session baseline, then the source session in room 14, then the post-source-session measurement. She needed rooms 13 and 15 to follow their regular schedule without interruptions.
She added to the protocol: confirm room 13 and 15 schedules through end of March. Ask coordinator whether schedule changes are notified in advance.
Then she thought about whether the first experiment's data was still usable.
It was, she decided, with appropriate caveats. The first and second baselines were not measuring the same thing — one was a post-session residual measurement and one was a during-session active measurement. This was a confound in the original protocol and a variable in the revised one. If she treated the two baselines as data points in the adjacent-room comparison rather than as checks on a stable baseline, they became evidence rather than noise.
She added: reframe first and second baselines as data points 1 and 2 in the adjacent-session comparison. Do not discard.
The measurement log was now four pages of protocol notes and three pages of data. She had come to room 14 to take a routine baseline measurement. She had found that the room was not doing what she expected it to do, and the thing it was doing instead was the most interesting possible version of the result.
She wrote at the bottom of the last page: the experiment is alive. This is what alive looks like.
Then she closed the notebook and made tea.